This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.

Bug 4978 - Is smlerr:output necessary?
Summary: Is smlerr:output necessary?
Alias: None
Product: SML
Classification: Unclassified
Component: Core (show other bugs)
Version: FPWD
Hardware: PC Windows XP
: P2 normal
Target Milestone: ---
Assignee: Valentina Popescu
QA Contact: SML Working Group discussion list
Keywords: editorial
: 4807 (view as bug list)
Depends on:
Blocks: 4647
  Show dependency treegraph
Reported: 2007-08-23 02:20 UTC by Virginia Smith
Modified: 2007-11-19 22:30 UTC (History)
1 user (show)

See Also:


Description Virginia Smith 2007-08-23 02:20:27 UTC
Regarding Section 5-Structured XML Output from Schematron Rules:

Does this section bring any significant value over the sch:diagnostic element? It seems that XML output, if needed, can be constructed as the user sees fit using sch:diagnostic element and the Schematron <name> and <value-of> elements.

In addition, ISO schematron specifies a Schematron Validation Report Language (SVRL) which is an XML 'report' file. See appendix D in the spec.

Can we drop section 5 or else work within the above 2 features provided by Schematron?
Comment 1 Virginia Smith 2007-10-03 00:07:16 UTC
*** Bug 4807 has been marked as a duplicate of this bug. ***
Comment 2 Kumar Pandit 2007-10-03 02:15:14 UTC
Since bug 4807 has been marked duplicate of this bug, I am copying below the proposal I had added to that bug. I have switched points #1 & #2 for clarifying the emphasis on pt #1.

I propose that we should remove this section entirely rather than spending time
fixing/clarifying it. There are a couple of reasons: 

1.       It is strange that the SML spec even attempts to define this. This
should either be implementation dependent or should be the responsibility of
the schematron spec to define. 

2.       Support for structured xml output is optional. It will be more
productive to spend our time on core (non-optional) SML issues. If we have lot
of time left after fixing all core issues, we can discuss whether to  bring
this back. 
Comment 3 Kumar Pandit 2007-10-04 05:01:07 UTC
SML spec does not define structure of error messages even for SML validation messages.
Comment 4 Virginia Smith 2007-11-01 19:08:49 UTC
Pratul will investigate what the original pre-submission group had in mind. Valentina will investigate possible use cases for this.
Comment 5 Pratul Dublish 2007-11-12 20:37:13 UTC
The primary scenario for this was impoverished devices that were incapable of doing SML model validation but wanted to perform some actions based on validation errors. An example is a management controller card that enforces a desired configuration on a computer. The desired config can be captured as an SML model, the controller card can discover the current state of the machine and send it to a remote server for validation against the SML model, get the structure error output, and use it to peform corrective actions to bring the computer's state in compliance with the desired config. It was felt that defining a mechanism, albeit optional,  in the SML spec will encourage vendors to adopt and imoplement it. 

Note that the above scenario is partially addressed by smlerr:output since it assumes that all desired configuration will be captured through Schematron rules. If SML identity constraints and target* constraints are used, then there is no standard mechanism for reporting these errors. 
Comment 6 Valentina Popescu 2007-11-15 19:14:58 UTC
I have investigated the usage and added my comments under this note

In summary : 
The structured XML output was defined to enable variable substitution support for localized messages, with the assumption that variable substitution cannot be achieved any other way.

In the proposal I submitted for , (see I show that there is possible to have variable substitution support by just using xsl:variable. As a result, I think this structured output is not necessary and can be removed from the spec.
Comment 7 Pratul Dublish 2007-11-15 19:19:02 UTC
As a member of the private WG that submitted SML to W3C, I disagree with the following assertion in Comment #6
In summary : 
The structured XML output was defined to enable variable substitution support
for localized messages, with the assumption that variable substitution cannot
be achieved any other way.

The primary scenario for this feature was the one listed in Comment #5. Localized error message support was added later, but it was not the primary driving scenario.
Comment 8 Valentina Popescu 2007-11-15 21:30:57 UTC
re comment #7

I based my comments on discussions I had with Harm Sluiman, one of the initial members of the workgroup, who had contributed to this section. I had an open action from the current group to investigate ( and by that clarify ) the usage of the XML structured output; since you were part of the initial group, which I was not aware of until now, I think it should have been easier if you owned that action..

So based on the results of my investigation which concluded that the XML output format was mainly intended for localization support, I concurred with the current proposal on removing this section as not required.
As more usescases are being unveiled, usecases that contradict my findings, I am not sure if the removal proposal is still acceptable.

Another think I decided to not pursue investigating due to obvious time constraints, is how this xml output format relates to the schematron validation report language - 

I assume the initial group had investigate this support and decided to create a new output under the SML spec for a valid reason. 
Comment 9 Pratul Dublish 2007-11-15 21:49:38 UTC
Sorry, I was just pointing out the primary use case that drove this feature in the private WG. As I pointed out in #5, structured error output is at best a partial solution for the primary scenario.  I agree with the recommendation in #6 (which is inline with the last straw poll we did on this issue) that smlerr:output should be removed from the spec. 

Comment 10 Pratul Dublish 2007-11-19 21:10:00 UTC
Please remove the smlerr:output section from the SML spec - this was agreed to on the 11/19 call
Comment 11 Valentina Popescu 2007-11-19 22:30:30 UTC
fixed as per comment #10

Section removed from SML spec:
6. Structured XML Output from Schematron Rules

sml-err-source.xsd modified to take out Structured XML Output types